Heritability of Autism Spectrum Disorder in a UK Population-Based Twin Sample

Research

Original Investigation Heritability of Autism Spectrum Disorder in a UK Population-Based Twin Sample

Emma Colvert, PhD; Beata Tick, MSc; Fiona McEwen, PhD; Catherine Stewart, PhD; Sarah R. Curran, MD; Emma Woodhouse, BSc; Nicola Gillan, BSc; Victoria Hallett, PhD; Stephanie Lietz, PhD; Tracy Garnett, DClinPsych; Angelica Ronald, PhD; Robert Plomin, PhD; Frühling Rijsdijk, PhD; Francesca Happé, PhD; Patrick Bolton, MD

Supplemental content at IMPORTANCE Most evidence to date highlights the importance of genetic influences on the jamapsychiatry.com liability to autism and related traits. However, most of these findings are derived from clinically ascertained samples, possibly missing individuals with subtler manifestations, and obtained estimates may not be representative of the population.

OBJECTIVES To establish the relative contributions of genetic and environmental factors in liability to autism spectrum disorder (ASD) and a broader autism phenotype in a large population-based twin sample and to ascertain the genetic/environmental relationship between dimensional trait measures and categorical diagnostic constructs of ASD.

DESIGN, SETTING, AND PARTICIPANTS We used data from the population-based cohort Twins Early Development Study, which included all twin pairs born in England and Wales from January 1, 1994, through December 31, 1996. We performed joint continuous-ordinal liability threshold model fitting using the full information maximum likelihood method to estimate genetic and environmental parameters of covariance. Twin pairs underwent the following assessments: the Childhood Autism Spectrum Test (CAST) (6423 pairs; mean age, 7.9 years), the Development and Well-being Assessment (DAWBA) (359 pairs; mean age, 10.3 years), the Autism Diagnostic Observation Schedule (ADOS) (203 pairs; mean age, 13.2 years), the Autism Diagnostic Interview–Revised (ADI-R) (205 pairs; mean age, 13.2 years), and a best-estimate diagnosis (207 pairs).

MAIN OUTCOMES AND MEASURES Participants underwent screening using a population-based measure of autistic traits (CAST assessment), structured diagnostic assessments (DAWBA, ADI-R, and ADOS), and a best-estimate diagnosis.

RESULTS On all ASD measures, correlations among monozygotic twins (range, 0.77-0.99) were significantly higher than those for dizygotic twins (range, 0.22-0.65), giving heritability estimates of 56% to 95%. The covariance of CAST and ASD diagnostic status (DAWBA, ADOS and best-estimate diagnosis) was largely explained by additive genetic factors (76%-95%). For the ADI-R only, shared environmental influences were significant (30% [95% CI, 8%-47%]) but smaller than genetic influences (56% [95% CI, 37%-82%]).

CONCLUSIONS AND RELEVANCE The liability to ASD and a more broadly defined high-level autism trait phenotype in this large population-based twin sample derives primarily from additive genetic and, to a lesser extent, nonshared environmental effects. The largely consistent results across different diagnostic tools suggest that the results are generalizable across multiple measures and assessment methods. Genetic factors underpinning individual differences in autismlike traits show considerable overlap with genetic influences on

diagnosed ASD. Author Affiliations: Author affiliations are listed at the end of this article. Corresponding Author: Beata Tick, MSc, MRC Social, Genetic, and Developmental Psychiatry Centre, Institute of Psychiatry, Psychology, and Neuroscience, King’s College London, De Crespigny Park, London JAMA Psychiatry. 2015;72(5):415-423. doi:10.1001/jamapsychiatry.2014.3028 SE5 8AF, England Published online March 4, 2015. ([email protected]).

(Reprinted) 415

Downloaded From: https://jamanetwork.com/ on 10/02/2021 Research Original Investigation Heritability of Autism Spectrum Disorder

win studies of autism,1-6 conducted from 1977 onward, analytic methods. Previous studies15,16 have identified their provided the first clear evidence that genetic factors twin samples through clinical services. Such a strategy could T were etiologically important. Recent reviews of this result in sampling bias; if registration or participation is influ- literature5,7-9 show general agreement across studies that con- enced by concordance, probandwise concordance rates in DZ cordance for autism in monozygotic (MZ) twin pairs is typi- twin pairs might be increased, resulting in inflated estimates cally at least double that in dizygotic (DZ) twin pairs, result- of common environmental influence. In addition, sole reli- ing in high heritability estimates (60%-90%)10-14 and suggesting ance on clinical ascertainment could result in underidentifi- little influence of shared environmental factors. Two twin cation of cases with high levels of functioning.32 Investiga- studies15,16 stand in contrast and reported only moderate heri- tors should include these cases and define the genetic liability tability (21%-38%), with a substantial shared environmental as a continuous distribution that extends beyond stringent di- component explaining 58% to 78% of the variance in liability agnostic categories. Our study is novel in using criterion- to autism spectrum disorder (ASD). In comparison, 1 recent standard, in-person, clinical diagnostic tools with a population- twin study did not confirm significant shared environmental based (vs a clinic-based) sample in which ascertainment was effects and reported heritability of 95%.17 In addition, a large good (62.1% response rate from the eligible sample compared population study of extended families (approximately 2 mil- with 17% in the study by Hallmayer et al15). lion individuals)18 reported estimates of 50% for heritability In summary, our first aim was to estimate heritability of and nonshared environmental factors. Most recently, in the the liability to ASD using a population-based sample that was same population, molecular genetic analysis19 indicated that selected using several screening instruments sent to all twins 95% of variance in ASD is accounted for by common allelic vari- in a 3-year birth cohort. The second aim was to study the ge- ants, supporting a polygenic model. This finding contrasts netic/environmental relationship between dimensional trait markedly with heritability estimates of around 0 derived from measures and categorical diagnostic constructs of ASD. In con- single-nucleotide polymorphism data (GCTA) in an arguably trast to previous approaches,15,18 we assumed a continuous li- underpowered sample.20 Given the interest in possible envi- ability distribution underlying ASD and a more broadly de- ronmental factors in the etiology of autism, these contradic- fined phenotype with high-level autism traits that fell short tory findings have reopened the discussion of high heritabil- of thresholds for an ASD diagnosis. We predicted a strong ge- ity and the possibility that findings may be biased by sample netic overlap between dimensional and diagnostic measures selection and screening. The first aim of the present study was in keeping with previous twin analyses based on extreme to examine the relative importance of genetic and environ- cases.21-23 mental factors in liability to ASD in a large systematically screened, population-based twin sample. 21,22 Twin and family studies have also shown that the ge- Methods netic liability to autism confers a risk for a broader range of im- pairments in social communication, restricted and repetitive Participants behaviors, and behaviors that extend beyond the traditional The participants were recruited from the Twins Early Devel- diagnostic boundaries for autism.9,23,24 These pioneering stud- opment Study,33 a longitudinal study of twin pairs ascer- ies contributed to the revision and broadening of diagnostic tained from population records of twin births in England and criteria and to the conceptualization of autism as a spectrum Wales from January 1, 1994, through December 31, 1996. The encompassing subtypes of pervasive developmental disor- Twins Early Development Study sample is considered repre- ders, such as Asperger syndrome, atypical autism, and sub- sentative of the population of the United Kingdom in terms of tler presentations.25,26 Research27-31 has explored autismlike maternal ethnicity (92.8% white) and educational level (40.1% traits in community samples and provided evidence of a ge- with A level qualifications or higher, the equivalent of some netic correlation between autismlike traits at the extremes and college education in the United States). Ethical authoriza- in the rest of the population. Our second aim was therefore to tion, including authorization to work with children, was given quantify the genetic and environmental relationship be- by the Institute of Psychiatry ethics committee. Parents were tween dimensional trait measures and the categorical diag- given a letter describing the general purpose of the study, and nostic constructs of ASD (from criterion-standard instru- written parental consent was required. Participation was vol- ments), which, to our knowledge, has not been tried before. untary and participants could withdraw from the study when- To provide a more definitive picture that addresses these ever they wished. 2 aims and incorporates current diagnostic concepts, we used The ASD and co-twin sample were selected after a 2-stage rigorous approaches and screened an age-specific epidemio- screening process outlined in Figure 1 and section 1 of eAp- logic sample of twins to ascertain all twins with possible ASD. pendix 1 in the Supplement. Of the 412 eligible families for the We then undertook independent, in-depth evaluations using Social Relationship Study at stage 1, 80.1% completed the De- an additional screening instrument and well-established di- velopment and Well-being Assessment (DAWBA) interview.34,35 agnostic assessment tools. The purpose was to minimize meth- At stage 2, 62.1% of the 235 eligible families underwent diag- odological artifacts and provide results that can be used as a nostic evaluations. Two researchers worked with each fam- benchmark for comparison in future research. ily. One researcher administered the Autism Diagnostic Inter- Our approach is novel and contrasts with those of other view–Revised (ADI-R)36 and the other administered the Autism recent twin and family studies in sample ascertainment and Diagnostic Observation Schedule (ADOS)37 for the first twin;

416 JAMA Psychiatry May 2015 Volume 72, Number 5 (Reprinted) jamapsychiatry.com

Downloaded From: https://jamanetwork.com/ on 10/02/2021 Heritability of Autism Spectrum Disorder Original Investigation Research

Figure 1. Social Relationship Study (SRS) Sample Selection Stages and the Overall Number of Participants Included in the Analysis

8941 TEDS families undergo screening

Screening stage 1 412 At-risk families selecteda 289 Families with ≥1 twin scoring ≥15 points on the CAST 210 Families with ≥1 twin reported to have existing ASD diagnosis

82 Families refused or no response (19.9%)

330 Families completed DAWBA telephone interviews

230 Families with ≥1 child who met 5 Families met eligibility criteria for DAWBA criteria for ASD TEDS but not recruited during earlier phases of study (identified via UK psychiatrists mail out and advertisement)

Screening stage 2 235 Families invited to take part in SRS

89 Families refused or no response (37.8%)

Face-to-face assessment 146 Families eligible for ASD home assessment

17 Families excluded from analysis (completed questionnaire booklet only) (11.6%)

80 Low-risk (CAST score, <12) 129 Families underwent home comparison families underwent assessment (1 excluded) home assessment (1 excluded)

29 Completed DAWBA

Twin analysis 359 Pairs included in DAWBA analysis 207 Pairs included best-estimate 91 Included in MZ pairs diagnosis analysis 133 Included in DZ SS pairs 56 Included in MZ pairs 135 Included in DZ OS pairs 77 Included in DZ SS pairs 74 Included in DZ OS pairs 203 Pairs included in ADOS analysis 55 Included in MZ pairs 77 Included in DZ SS pairs 71 Included in DZ OS pairs 205 Pairs included in ADI-R analysis 56 Included in MZ pairs 76 Included in DZ SS pairs 73 Included in DZ OS pairs

ADI-R indicates Autism Diagnostic Interview–Revised; ADOS, Autism Diagnostic Observation Schedule; ASD, autism spectrum disorder; DAWBA, Development and Well-being Assessment; DZ, dizygotic; MZ, monozygotic; OS, opposite-sex; SS, same-sex; and TEDS, Twins Early Development Study. a Some twins met both criteria, so numbers total more than 412.

2 the reseachers then swapped assessments for the second twin. suspected ASD) but who did not take part (zygosity, χ 1 = 1.5

This design meant that different assessors administered the [P = .23]; socioeconomic status, t397 =−1.2[P = .25, indepen- ADI-R and ADOS assessments within each pair to minimize any dent t test with 2-tailed significance]; and CAST result,

effects of rater bias. t420 =−1.5[P = .14, independent t test with 2-tailed signifi- 2 Within the final sample, participants with ASD were cance]), with the exception of sex (χ1 =20.1[P < .001]). Among broadly comparable to those eligible for participation (score the group with high CAST scores or suspected ASD, 36.4% were of ≥15 on the Childhood Autism Spectrum Test [CAST]25 or with female compared with 16.6% of the final sample.

jamapsychiatry.com (Reprinted) JAMA Psychiatry May 2015 Volume 72, Number 5 417

Downloaded From: https://jamanetwork.com/ on 10/02/2021 Research Original Investigation Heritability of Autism Spectrum Disorder

Measures iors and interests. Clinical cutoffs were available for ASD and Childhood Autism Spectrum Test autism, and these diagnostic groups were combined to create The CAST is an informant-completed questionnaire based on a single ASD category. An additional broad-spectrum cat- behavioral descriptions of ASD as delineated in the Interna- egory included individuals who scored just below the cutoff tional Statistical Classification of Diseases, Tenth Revision (ICD- (−2 points) to correspond to the broad-spectrum category on 10) and the DSM-IV. The 31 items are scored yes or no and the ADI-R. The measure used in the analysis was a 3-category summed; a cutoff score of 15 or greater is reported to have 100% diagnosis of ASD, where 0 indicates no ASD/controls; 1, broad- sensitivity, 97% specificity, and a positive predictive value of spectrum disorder; and 2, ASD. 50% for a diagnosis of ASD.28 The CAST data from at least 1 twin were available from 6423 pairs (MZ pairs, 2261; DZ same-sex Best-Estimate Diagnosis pairs, 2097; and DZ opposite-sex pairs, 2065), with a mean (SD) Diagnoses were assigned with investigators blinded to zygos- age of 7.9 (0.5) years. Of these, 289 pairs (4.5% of all pairs; 317 ity and co-twin diagnostic status after review of all available individuals) had scores greater than the cutoff value. information (ADI-R, ADOS, and DAWBA assessments and clinical reports). When all available sources of information were in Development and Well-being Assessment agreement, cases were assigned to that category. In 89 cases, Telephone interviews using the ASD module of the DAWBA the diagnostic classifications across instruments were incon- were performed at the second stage35 and included 15 ques- sistent. In these cases, all available data were assessed by ex- tions about social difficulties; 14 questions about repetitive, pert clinicians (E.C., S.R.C., and/or P.B.), and best-estimate di- restricted behaviors and interests; and 3 questions about de- agnoses (BeDs) were assigned based on this review. Additional velopmental language milestones. The same parent rated both details are given in eAppendix 2 in the Supplement. Best- twins during a telephone call with a single interviewer. A child estimate diagnosis was used in the analysis as a 3-category mea- received a DAWBA diagnosis of autism when the operational sure of ASD, where 0 indicates no ASD/controls; 1, broad- criteria in the DSM-IV and ICD-10 were met. A diagnosis of As- spectrum disorder; and 2, ASD. perger syndrome was given when parent reports indicated that all autism criteria were met but the child’s early language de- Data Analysis velopment was not delayed and the child’s intellectual ability Twin Correlations was in the normal range. A diagnosis of ASD (other) was as- Twin data analysis was performed in the structural equation signed if the parents reported a minimum of 3 probable or 2 modeling program OpenMx.38 We used the full information definite symptoms from the social difficulties domain, 2 prob- maximum likelihood estimation to jointly analyze the con- able or 1 definite symptom from the communication domain, tinuous (CAST) and ordinal (ASD) measures, assuming a liabil- and 2 probable or 1 definite symptom from the repetitive, re- ity threshold model to reflect the risk for ASD.39,40 To obtain stricted behaviors and interests domain. The measure used in unbiased estimates, thresholds were fixed to z values corre- analysis was a 3-category diagnosis of ASD, where 0 indicates sponding to the following known population prevalences be- no ASD/controls; 1, ASD (other); and 2, Asperger syndrome or cause of the selection of individuals with ASD: first at 5%,41,42 autism. discriminating between unaffected and broad-spectrum disorder, and second at 1%, discriminating between broad- Autism Diagnostic Interview–Revised spectrum disorder and ASD. The assumption of a joint multi- The ADI-R is a well-established diagnostic tool for the assess- variate normal distribution for CAST scores and the 3 ASD ment of autism.36 It consists of a semistructured caregiver in- diagnostic categories (unaffected, broad-spectrum disorder, terview inquiring about current function and developmental his- and ASD) allowed the estimation of correlations within and tory (93 items) and is administered by a trained investigator. We across MZ/DZ twin pairs. The MZ:DZ ratios of these correla- used criteria from the Autism Genetics Resource Exchange (http: tions indicate the relative importance of genetic and environ- //agre.autismspeaks.org/site/c.lwLZKnN1LtH/b.5332889/k.B473 mental influences on variation within each measure and on the /AGRE.htm) to assign cases to 1 of the following 3 categories: ASD covariance between them; these correlations were formally (consisting of Autism Genetics Resource Exchange categories tested in the bivariate genetic model. autism and not quite autism), broad-spectrum disorder, and unaffected (operational definitions are given in eAppendix 2 in the The Bivariate Genetic Model Supplement). The measure used in the analysis was a 3-cat- With the use of biometrical genetic theory, the covariance of egory diagnosis of ASD, where 0 indicates no ASD/controls; 1, the CAST score and each ASD diagnosis was modeled as the broad-spectrum disorder; and 2, ASD. effects of additive genetic (A), shared environmental (C), and nonshared environmental and measurement error (E) factors.43 Autism Diagnostic Observation Schedule Because the order of traits is immaterial, we interpreted the

The ADOS is a well-validated, semistructured observational as- standardized solution in which the paths from the A1 factor to

sessment designed to accompany the ADI-R in the diagnosis the CAST score and the A2 factor to ASD are the square roots of ASD.37 The present study used recent updates to the ADOS of their respective heritabilities, and the correlation between 44,45 algorithm (Catherine Lord, PhD, written communication, 2008; A1 and A2 is the genetic correlation between them (rA). The described in eAppendix 2 in the Supplement) to yield scores same logic applies to the nonshared environmental effects for communication, social interaction, and restricted behav- (Figure 2). Shared environmental factors were modeled on ASD

418 JAMA Psychiatry May 2015 Volume 72, Number 5 (Reprinted) jamapsychiatry.com

Downloaded From: https://jamanetwork.com/ on 10/02/2021 Heritability of Autism Spectrum Disorder Original Investigation Research

Figure 2. Diagram of the Correlated Factors Solution of the Joint Table 1. MZ and DZ Probandwise Concordance Rates Continuous-Ordinal Model of Effects on the Childhood Autism Spectrum Across MZ and DZ Affected Twinsa Test (CAST) Results and Best-Estimate Diagnosis (BeD) MZ Twin Pairs DZ Twin Pairs No. of No. of r A = 0.70 Discordant/ Probandwise Discordant/ Probandwise (0.63-0.80) Concordant Concordance Concordant Concordance A1 A2 C2 Measure Pairs Rate Pairs Rate ASDb √0.78 √0.95 √0.00 DAWBA 12/15 0.71 74/2 0.05 (0.77-0.79) (0.78-0.98) (0.00-0.26) ADI-R 15/12 0.62 80/8 0.17 ADOS 8/12 0.75 57/9 0.40 CAST Best-estimate diagnosis Best- 8/17 0.87 77/11 0.22 estimate diagnosis √0.22 √0.05 (0.21-0.23) (0.02-0.17) ASD and Broad-Spectrum Disorderc DAWBA 16/24 0.75 118/5 0.08

E1 E2 ADI-R 4/24 0.92 54/43 0.61 r E = 0.48 ADOS 7/16 0.82 56/18 0.39 (0.16-0.84) Best- 3/24 0.94 70/30 0.46 estimate diagnosis The model represents the standardized effects of additive genetic (A1 and A2) and nonshared environmental factors (E1 and E2) on each trait separately and Abbreviations: ADI-R, Autism Diagnostic Interview–Revised; ADOS, Autism the A and E correlations between the 2 variables (rA and rE). We found no Diagnostic Observation Schedule; ASD, autism spectrum disorder; shared environmental influences at play for the CAST score; therefore, no DAWBA, Development and Well-being Assessment; DZ, dizygotic; covariance due to correlated shared environmental factors is possible with the MZ, monozygotic. autism spectrum disorder (ASD) measures. We modeled the shared a Includes same-sex and opposite-sex twin pairs. environmental effects for the ASD measures, which were nonsignificant for b most as shown for the BeD (striped line). Rates reflect twins included in category 2 (ASD) only. c Rates reflect pairs in which a child was included in category 1 (broad-spectrum disorder) or 2.

only; they do not influence the variance of the CAST scores46 and therefore cannot explain the covariance with ASD. In ad- definedweightsusedsothatthe0-2cellsgetafullweightof1 dition to the standardized path estimates, we calculated the and the 0-1 and 1-2 cells only a weight of 25% in calculating dis- ͙ 2 phenotypic correlation (rph) due to genetic effects (rph_A)as h1 agreement. These values (−1 to 1) represent the observed agree- ͙ 2 × rA × h2 and the phenotypic correlation due to nonshared ment between 2 diagnostic tests relative to the expected agree- ͙ 2 49 environmental and measurement error effects (rph_E)as e1 ment between tests occurring by chance alone. We found ͙ 2 × rE × e2 , which can be expressed as proportions of pheno- moderate κ agreement for DAWBA and the ADOS assessment typic correlation.47,48 (0.58) and substantial agreement for DAWBA and the ADI-R assessment and for DAWBA and BeD (both 0.72), for the ADI-R and ADOS assessments (0.67), and for the ADOS assessment Results and BeD (0.79). Agreement for the ADI-R assessment and BeD was almost perfect (0.91). Probandwise Concordance Rates Probandwise concordance rates were calculated as [2 × Twin Correlations (number of concordant pairs)]/[2 × (number of concordant The 2:1 MZ:DZ ratio of the cross-twin within-trait correla- pairs) + discordant pairs] (Table 1). These calculations ex- tions for the ADOS assessment and BeD suggest a significant press the probability that the co-twin of a proband (affected contribution of genetic effects, with the remainder explained twin) is also affected and are commonly used as an index of by nonshared environmental effects (Table 2). This contribu- twin resemblance. The high MZ (0.62-0.94) and low DZ (0.05- tion is not the case for DAWBA, for which the DZ correlation is 0.61) concordances suggest substantial genetic influence. For less than half that of the MZ pairs, pointing to nonadditive ge- example, MZ concordances are 0.87 for ASD and 0.94 for BeD, netic effects. For the ADI-R assessment, the DZ correlation is in contrast to 0.22 and 0.46, respectively, for DZ concor- more than half the MZ correlation, indicating genetic and dances. However, concordance rates cannot be used to esti- shared environmental effects. The MZ:DZ ratio of the cross- mate genetic and environmental parameters because they do twin cross-trait correlations for the CAST score and each di- not take population prevalence rates into account. agnosis indicates mainly genetic and nonshared environmental influences on their overlap. Diagnostic Agreement Agreement of classification of individuals into the 3 catego- Bivariate Genetic Model ries (unaffected, broad-spectrum disorder, and ASD) for dif- Table 3 reports the standardized results of the bivariate ACE ferent diagnostic measures was calculated by means of models. Variance in the CAST score (age and sex regressed) re- weighted κ coefficients (Stata software; StataCorp) with pre- sulted from genetic influences (78% [95% CI, 77%-79%]) and

jamapsychiatry.com (Reprinted) JAMA Psychiatry May 2015 Volume 72, Number 5 419

Downloaded From: https://jamanetwork.com/ on 10/02/2021 Research Original Investigation Heritability of Autism Spectrum Disorder

Table 2. MZ and DZ Within-Trait and Cross-Trait Twin Correlationsa

Within-Trait Correlation, r (95% CI)b Cross-Twin Correlation, r (95% CI)c Measure MZ DZ MZ Cross-Trait DZ Cross-Trait CAST 0.79 (0.77-0.80)d,e 0.31 (0.28-0.33)d,e NA NA DAWBA 0.82 (0.67-0.90)e 0.22 (0.00-0.42) 0.40 (0.34-0.45)e −0.01 (0.00-0.07) ADI-R 0.87 (0.77-0.93)e 0.65 (0.55-0.73)e 0.60 (0.54-0.65)e 0.40 (0.35-0.45)e ADOS 0.77 (0.62-0.87)e 0.40 (0.26-0.63)e 0.56 (0.49-0.61)e 0.30 (0.23-0.37)e Best-estimate diagnosis 0.99 (0.98-0.99)e 0.53 (0.41-0.63)e 0.61 (0.57-0.66)e 0.37 (0.31-0.42)e

Abbreviations: See Table 1. CAST, Covariance of the Childhood Autism Spectrum c For the CAST score, 4 sets of correlations are available because 4 bivariate Test; NA, not applicable. analyses were performed; here, only 1 is given (the other 3 were of similar a Based on 4 bivariate analyses of the CAST score with each diagnostic ASD value and with overlapping 95% CIs). measure. d Maximum likelihood cross-twin, cross-CAST correlation was obtained for each b Maximum likelihood correlations for MZ and DZ twins (including same-sex and diagnostic variable and the CAST score separately. opposite-sex DZ pairs) are estimated in a model with the 2 thresholds on the e Indicates significant estimates (ie, 95% CIs not spanning 0). liability to ASD set to population values of broad-spectrum disorder (5%) and ASD (1%) prevalence.

Table 3. Standardized Estimates of the Reduced A(C)E Bivariate Models of CAST Scores and Each of the 4 Clinical Measures of ASDa

Standardized Estimates (95% CI) Part of Phenotypic Correlation (%) 2 2 2 b b c d d Measure h c e rA rE rph rph_A rph_E CAST 0.78 0.00 0.22 NA NA NA NA NA (0.77-0.79)e (0.00-0.00) (0.21-0.23)e DAWBA 0.78 0.00 0.22 0.52 0.48 0.52 0.40 (77)e 0.12 (23)e (0.48-0.87)e (0.00-0.00) (0.13-0.36)e (0.47-0.67)e (0.32-0.64)e (0.48-0.55)e ADI-R 0.56 0.30 0.14 0.89 0.19 0.61 0.58 (≈100)e 0.03 (≈0) (0.37-0.82)e (0.08-0.47)e (0.07-0.47)e (0.70-0.99)e (0.00-0.41) (0.56-0.66)e ADOS 0.76 0.00 0.24 0.73 −0.02 0.54 0.56 (≈100)e −0.02 (≈0) (0.41-0.86)e (0.00-0.30) (0.14-0.39)e (0.63-0.99)e (0.00-0.15) (0.51-0.60)e Best-estimate 0.95 0.00 0.05 0.70 0.48 0.65 0.60 (92)e 0.05 (8)e diagnosis (0.74-0.98)e (0.00-0.26) (0.02-0.17)e (0.63-0.80)e (0.16-0.84)e (0.24-0.67)e Abbreviations: See Table 1. c2, shared environmental factors estimate; CAST, d Indicates part of the phenotypic correlation due to additive genetic (A) and Covariance of the Childhood Autism Spectrum Test; e2, nonshared unique environmental (E) influences (percentage of phenotypic correlation). environmental factors estimate; h2, heritability estimate; NA, not applicable. Proportions cannot be calculated for the ADOS assessment owing to the a Thresholds on the ASD liability were fixed at 5% (broad spectrum) and 1% opposite signs of rph_A and rph_E; however, if we disregard the nonsignificant (ASD); the estimates for the CAST score across the 4 models were of similar contributions of rph_E for the CAST-ADOS and CAST–ADI-R relationships, value and with overlapping 95% CIs. shared genetic effects explain nearly all of the observed correlations. e b Indicates additive genetic (A) and nonshared environmental (E) correlations Indicates significant estimates (ie, 95% CIs not spanning 0). between the CAST score and ASD measure. c Indicates phenotypic (ph) correlation between the CAST score and ASD measures.

nonshared environmental effects (22% [95% CI, 21%-23%]), as dardized estimates of the reduced bivariate AE model (in which reported previously.30 Genetic influences were significant for no shared environmental influences are found) for the CAST all clinical measures with the highest heritability reported for score and BeD, BeD being the best diagnostic estimate of ASD BeD (95% [95% CI, 74%-98%]) and shared environments sig- in our study. The findings indicate strong and overlapping ge- nificantly explaining the variance of the ADI-R assessment only netic influences on dimensional and categorical measures. (30% [95% CI, 8%-47%]). The correlations between the CAST

score and each of the ASD variables (rph) is the sum of the paths via the additive genetic and nonshared environmental fac- Discussion tors (A and E) connecting the 2 variables (rph_A and rph_E). The phenotypic correlations were moderate to high (0.52-0.65), and The present study examines the genetic and environmental genetic factors accounted for 77% to 100% of the covariance. contributions to ASD in a large systematically screened popu- The genetic correlations, that is, the extent to which the same lation-based twin sample and the genetic/environmental over- genetic factors influence the CAST score and clinical mea- lap between a continuous measure of autismlike traits and cat- sures independent of their heritabilities, are substantial (0.52- egorical diagnostic assessments. Our study was novel in 0.89). The remainder of the covariance was explained by non- including twins with high subclinical levels of traits and se-

shared environmental factors (rph_E), although nonsignificant lected low-risk twins as well as those with diagnosed ASD to for the overlap between the CAST score and the ADI-R or the capture the full range of liability. The probandwise concor- ADOS assessment. Figure 2 depicts the path diagram with stan- dance rates and liability threshold model analyses reassert the

420 JAMA Psychiatry May 2015 Volume 72, Number 5 (Reprinted) jamapsychiatry.com

Downloaded From: https://jamanetwork.com/ on 10/02/2021 Heritability of Autism Spectrum Disorder Original Investigation Research

importance of genetic factors in the etiology of ASD. Analyses picture of genetic risk (additive and nonadditive) for ASD than partitioning liability into genetic and shared and nonshared en- in previous studies. A novel contribution is the strong evi- vironmental components indicate that most liability could be dence that the same genetic influences are largely respon- attributed to additive genetic influences and a smaller propor- sible for the overlap between dimensional trait measures and tion attributed to nonshared environmental influences. This categorical diagnostic constructs of ASD. In addition, to the best finding held across a number of different measures. We found of our knowledge, this study is one of the largest screened very little evidence of shared environmental effects overall, population-based twin studies yet reported. which is contrary to the findings of Hallmayer et al,15 although The limitations of this study include the fact that few of the wide CIs in their results for additive genetic and shared en- the potentially eligible twin pairs did not enroll in the study. vironmental effects overlap with some of the present esti- Second, although one of the largest twin studies, the sample mates. In our study, only the ADI-R parent-reported develop- size was insufficient to allow any meaningful analyses of the mental history measure showed significant shared environment basis for sex differences in ASD. In addition, twin study meth- effects. Because the ADI-R assessment was completed by the ods assume that the environments of MZ and DZ53 twins are same parent for both twins, the estimated influence of shared equal and that twins are not at especially high risk for the dis- environment may be inflated by rater bias. However, the wide order under investigation. The available evidence indicates that CIs (0.08-0.47) warrant caution in interpretation. both assumptions are justified in this study.54 Another issue Our findings also confirm that the heritability of the liabil- is that genetic modeling assumes that no gene-environment ity to ASD is high when incorporating subclinical cases with interactions or correlations exist; if they exist, the estimates high trait scores into the model, extending support for the no- of environmental and genetic effects may be inflated.55 Heri- tion that the genetic liability to autism confers a risk for a tability estimates are also population specific and depend on broader autism phenotype. Indeed, the relationship between the dynamic interaction with the current environment. Our the CAST score and diagnostic assessments indicated a sub- analysis took a liability threshold approach, but other types of stantial genetic correlation and a significant correlation in the analyses (eg, continuous data modeling, DeFries-Fulker quan- nonshared environmental factors that influence variations in tile regression56) are possible and may be warranted by fu- both traits. This result indicates common etiologic underpin- ture developments in the molecular genetics of ASD. Recent nings for individual differences in autistic traits across the findings lend support to a polygenic trait approach.19 whole spectrum and in our 3 clinically meaningful categories (ASD, high subclinical trait level, and low risk/trait level). This result provides support for examining broader autistic traits Conclusions in the general population as a complementary strategy for iden- tifying the genetic risk factors for ASD.50-52 Our findings are The present study combines the strengths of previous stud- broadly in line with those of recent twin and family studies and ies and provides a more complete picture than any of them in- point toward strong genetic effects in ASD and no strong in- dividually by being nationally representative and incorporat- fluence from shared environmental factors. The strengths of ing dimensional and categorical measures using a systematic the present study add validity to these conclusions because repeated screening method. We conclude that liability to ASD previous research has often lacked the rigor and systematic ap- and a more broadly defined high-level autism trait pheno- proach to the sample selection used herein. The population- type in UK twins 8 years or older derives from substantial ge- based sampling in the present study, the 2-stage systematic netic and moderate nonshared environmental influences. Ge- screening methods used, and the inclusion of individuals with netic influences on diagnosed ASD are shared with those on subclinical disorders ensured the capture of a more complete autistic traits in the general population.

ARTICLE INFORMATION Psychology, and Neuroscience, King’s College Study concept and design: Colvert, Tick, Ronald, Submitted for Publication: July 9, 2014; final London, London, England (Stewart, Curran, Plomin, Rijsdijk, Happé, Bolton. revision received November 18, 2014; accepted Hallett); Brighton and Sussex Medical School, Acquisition, analysis, or interpretation of data: November 19, 2014. University of Sussex, East Sussex, England (Curran); Colvert, Tick, McEwen, Stewart, Curran, Sussex Partnership NHS Foundation Trust, Trust Woodhouse, Gillan, Hallett, Lietz, Garnett, Ronald, Published Online: March 4, 2015. Headquarters, West Sussex, England (Curran); Happé, Bolton. doi:10.1001/jamapsychiatry.2014.3028. Department of Forensic and Neurodevelopmental Drafting of the manuscript: Colvert, Tick, McEwen, Author Affiliations: MRC Social, Genetic, and Sciences, Institute of Psychiatry, Psychology, and Curran, Rijsdijk, Happé, Bolton. Developmental Psychiatry Centre, Institute of Neuroscience, King’s College London, London, Critical revision of the manuscript for important Psychiatry, Psychology, and Neuroscience, King’s England (Woodhouse); Research Department of intellectual content: Colvert, Tick, McEwen, College London, London, England (Colvert, Tick, Clinical, Educational and Health Psychology, Stewart, Woodhouse, Gillan, Hallett, Lietz, Garnett, McEwen, Ronald, Plomin, Rijsdijk, Happé, Bolton); University College London, London, England Ronald, Plomin, Rijsdijk, Happé, Bolton. Department of Child and Adolescent Psychiatry, (Lietz); Department of Psychological Sciences, Statistical analysis: Colvert, Tick, Ronald, Rijsdijk, Institute of Psychiatry, Psychology, and University of London, London, England (Ronald). Bolton. Neuroscience, King’s College London, London, Author Contributions: Dr Colvert and Ms Tick Obtained funding: Plomin, Happé, Bolton. England (McEwen, Bolton); South London and contributed equally to this article, as did Drs Happé Administrative, technical, or material support: Maudsley NHS (National Health Service) and Bolton. Ms Tick and Dr Rijsdijk had full access to Colvert, McEwen, Stewart, Woodhouse, Gillan, Foundation Trust, Maudsley Hospital, London, all the data in the study and take responsibility for Garnett, Ronald. England (Stewart, Gillan, Garnett, Bolton); the integrity of the data and the accuracy of the Study supervision: Curran, Rijsdijk, Happé, Bolton. Department of Psychology, Institute of Psychiatry, data analysis. Conflict of Interest Disclosures: None reported.

jamapsychiatry.com (Reprinted) JAMA Psychiatry May 2015 Volume 72, Number 5 421

Downloaded From: https://jamanetwork.com/ on 10/02/2021 Research Original Investigation Heritability of Autism Spectrum Disorder

Funding/Support: The Twins Early Development aetiologies of parent, teacher and child ratings of 29. Robinson EB, Koenen KC, McCormick MC, et al. Study (TEDS) is supported by program grant autistic-like traits and their overlap. Eur Child Evidence that autistic traits show the same etiology G0901245 (previously G0500079) from the UK Adolesc Psychiatry. 2008;17(8):473-483. in the general population and at the quantitative Medical Research Council (MRC). The Social 13. Skuse DH, Mandy WPL, Scourfield J. Measuring extremes (5%, 2.5%, and 1%). Arch Gen Psychiatry. Relationship Study was supported by grant autistic traits: heritability, reliability and validity of 2011;68(11):1113-1121. G0500870 from the MRC. This study is also the Social and Communication Disorders Checklist. 30. Ronald A, Happé F, Bolton P, et al. Genetic supported by 1+3 PhD studentship MR/J500380/1 Br J Psychiatry. 2005;187:568-572. heterogeneity between the three components of from the MRC (Ms Tick), by a senior investigator 14. Hoekstra RA, Bartels M, Verweij CJ, Boomsma the autism spectrum: a twin study. J Am Acad Child award from the National Institute for Health Adolesc Psychiatry. 2006;45(6):691-699. Research (Dr Bolton), by the Biomedical Research DI. Heritability of autistic traits in the general Centre in Mental Health at the South London and population. Arch Pediatr Adolesc Med. 2007;161(4): 31. Lundström S, Chang Z, Råstam M, et al. Autism Maudsley NHS Trust (Dr Bolton), and by an Autism 372-377. spectrum disorders and autistic like traits: similar Speaks grant (Dr Bolton). 15. Hallmayer J, Cleveland S, Torres A, et al. Genetic etiology in the extreme end and the normal variation. Arch Gen Psychiatry. 2012;69(1):46-52. Role of the Funder/Sponsor: The funding source heritability and shared environmental factors had no role in the design and conduct of the study; among twin pairs with autism. Arch Gen Psychiatry. 32. Baron-Cohen S, Scott FJ, Allison C, et al. collection, management, analysis, and 2011;68(11):1095-1102. Prevalence of autism-spectrum conditions: UK interpretation of the data; preparation, review, or 16. Frazier TW, Thompson L, Youngstrom EA, et al. school-based population study. Br J Psychiatry. approval of the manuscript; and decision to submit A twin study of heritable and shared environmental 2009;194(6):500-509. the manuscript for publication. contributions to autism. J Autism Dev Disord.2014; 33. Haworth CMA, Davis OSP, Plomin R. Twins Additional Contributions: We thank the TEDS 44(8):2013-2025. Early Development Study (TEDS): a genetically participants and their families for their ongoing 17. Nordenbæk C, Jørgensen M, Kyvik KO, sensitive investigation of cognitive and behavioral contribution. Bilenberg N. A Danish population-based twin study development from childhood to young adulthood. on autism spectrum disorders. Eur Child Adolesc Twin Res Hum Genet. 2013;16(1):117-125. REFERENCES Psychiatry. 2014;23(1):35-43. 34. Dworzynski K, Happé F, Bolton P, Ronald A. 1. Folstein S, Rutter M. Genetic influences and 18. Sandin S, Lichtenstein P, Kuja-Halkola R, Relationship between symptom domains in autism infantile autism. Nature. 1977;265(5596):726-728. Larsson H, Hultman CM, Reichenberg A. The spectrum disorders: a population based twin study. J Autism Dev Disord. 2009;39(8):1197-1210. 2. Folstein S, Rutter M. Infantile autism: a genetic familial risk of autism. JAMA. 2014;311(17):1770-1777. study of 21 twin pairs. J Child Psychol Psychiatry. 19. Gaugler T, Klei L, Sanders SJ, et al. Most genetic 35. Goodman R, Ford T, Richards H, Gatward R, 1977;18(4):297-321. risk for autism resides with common variation. Nat Meltzer H. The Development and Well-Being Genet. 2014;46(8):881-885. Assessment: description and initial validation of an 3. Ritvo ER, Freeman BJ, Mason-Brothers A, Mo integrated assessment of child and adolescent A, Ritvo AM. Concordance for the syndrome of 20. Trzaskowski M, Dale PS, Plomin R. No genetic psychopathology. J Child Psychol Psychiatry. 2000; autism in 40 pairs of afflicted twins. Am J Psychiatry. influence for childhood behavior problems from 41(5):645-655. 1985;142(1):74-77. DNA analysis. J Am Acad Child Adolesc Psychiatry. 36. Lord C, Rutter M, Le Couteur A. Autism 4. Steffenburg S, Gillberg C, Hellgren L, et al. A twin 2013;52(10):1048-1056.e3. doi:10.1016/j.jaac.2013 .07.016. Diagnostic Interview–Revised: a revised version of a study of autism in Denmark, Finland, Iceland, diagnostic interview for caregivers of individuals Norway and Sweden. J Child Psychol Psychiatry. 21. Bishop DVM, Maybery M, Wong D, Maley A, with possible pervasive developmental disorders. 1989;30(3):405-416. Hallmayer J. Characteristics of the broader J Autism Dev Disord. 1994;24(5):659-685. phenotype in autism: a study of siblings using the 5. Bailey A, Le Couteur A, Gottesman I, et al. 37. Lord C, Rutter M, Goode S, et al. Autism Autism as a strongly genetic disorder: evidence Children’s Communication Checklist-2. Am J Med Genet B Neuropsychiatr Genet. 2006;141B(2):117-122. Diagnostic Observation Schedule: a standardized from a British twin study. Psychol Med. 1995;25(1): observation of communicative and social behavior. 63-77. 22. Constantino JN, Todorov A, Hilton C, et al. J Autism Dev Disord. 1989;19(2):185-212. Autism recurrence in half siblings: strong support 6. Lichtenstein P, Carlström E, Råstam M, Gillberg 38. Boker S, Neale M, Maes H, et al. OpenMx: an C, Anckarsäter H. The genetics of autism spectrum for genetic mechanisms of transmission in ASD. Mol Psychiatry. 2013;18(2):137-138. open source extended structural equation disorders and related neuropsychiatric disorders in modeling framework. Psychometrika. 2011;76(2): childhood. Am J Psychiatry. 2010;167(11):1357-1363. 23. Sucksmith E, Roth I, Hoekstra RA. Autistic traits 306-317. below the clinical threshold: re-examining the 7. Taniai H, Nishiyama T, Miyachi T, Imaeda M, Sumi 39. Falconer DS. The inheritance of liability to S. Genetic influences on the broad spectrum of broader autism phenotype in the 21st century. Neuropsychol Rev. 2011;21(4):360-389. certain diseases, estimated from the incidence autism: study of proband-ascertained twins. Am J among relatives. Ann Hum Genet. 1965;29(1):51-76. Med Genet B Neuropsychiatr Genet. 2008;147B(6): 24. Pickles A, Starr E, Kazak S, et al. Variable 844-849. expression of the autism broader phenotype: 40. Pearson K. On the laws of inheritance in man, II: on the inheritance of the mental and moral 8. Rosenberg RE, Daniels AM, Law JK, Law PA, findings from extended pedigrees. J Child Psychol Psychiatry. 2000;41(4):491-502. characters in man, and its comparison with the Kaufmann WE. Trends in autism spectrum disorder inheritance of the physical characters. Biometrika. diagnoses: 1994-2007. J Autism Dev Disord. 25. Williams JG, Allison C, Scott FJ, et al. The 1904;3(2/3):131-190. 2009;39(8):1099-1111. Childhood Autism Spectrum Test (CAST): sex 41. Brugha TS, McManus S, Bankart J, et al. 9. Ronald A, Hoekstra RA. Autism spectrum differences. J Autism Dev Disord. 2008;38(9):1731- 1739. Epidemiology of autism spectrum disorders in disorders and autistic traits: a decade of new twin adults in the community in England. Arch Gen studies. Am J Med Genet B Neuropsychiatr Genet. 26. Baron-Cohen S, Wheelwright S, Skinner R, Psychiatry. 2011;68(5):459-465. 2011;156B(3):255-274. Martin J, Clubley E. The Autism-Spectrum Quotient 42. Baird G, Simonoff E, Pickles A, et al. Prevalence 10. Ronald A, Happé F, Plomin R. The genetic (AQ): evidence from Asperger syndrome /high-functioning autism, males and females, of disorders of the autism spectrum in a population relationship between individual differences in social cohort of children in South Thames: the Special and nonsocial behaviours characteristic of autism. scientists and mathematicians. J Autism Dev Disord. 2001;31(1):5-17. Needs and Autism Project (SNAP). Lancet. 2006; Dev Sci. 2005;8(5):444-458. 368(9531):210-215. 27. Constantino JN. The quantitative nature of 11. Ronald A, Happé F, Price TS, Baron-Cohen S, 43. Neale MC, Cardon LR. Methodology for Genetic Plomin R. Phenotypic and genetic overlap between autistic social impairment. Pediatr Res. 2011;69(5, pt 2):55R-62R. Studies of Twins and Families. Dordrecht, the autistic traits at the extremes of the general Netherlands: Kluwer Academic Publishers; 1992. population. J Am Acad Child Adolesc Psychiatry. 28. Williams J, Scott F, Stott C, et al. The CAST 2006;45(10):1206-1214. (Childhood Asperger Syndrome Test): test accuracy. 44. Loehlin JC. The Cholesky approach: a cautionary note. Behav Genet. 1996;26(1):65-69. 12. Ronald A, Happé F, Plomin R. A twin study Autism. 2005;9(1):45-68. investigating the genetic and environmental

422 JAMA Psychiatry May 2015 Volume 72, Number 5 (Reprinted) jamapsychiatry.com

Downloaded From: https://jamanetwork.com/ on 10/02/2021 Heritability of Autism Spectrum Disorder Original Investigation Research

45. Rijsdijk FV, van Haren NE, Picchioni MM, et al. 49. Viera AJ, Garrett JM. Understanding childhood and adolescence. Mol Autism. 2014;5(1): Brain MRI abnormalities in schizophrenia: same interobserver agreement: the kappa statistic. Fam 18. doi:10.1186/2040-2392-5-18. genes or same environment? Psychol Med. 2005;35 Med. 2005;37(5):360-363. 53. Rijsdijk FV, Sham PC. Analytic approaches to (10):1399-1409. 50. Ronald A, Butcher LM, Docherty S, et al. twin data using structural equation models. Brief 46. Holmboe K, Rijsdijk FV, Hallett V, Happé F, A genome-wide association study of social and Bioinform. 2002;3(2):119-133. Plomin R, Ronald A. Strong genetic influences on non-social autistic-like traits in the general 54. Curran S, Dworzynski K, Happé F, et al. No the stability of autistic traits in childhood. J Am Acad population using pooled DNA, 500 K SNP major effect of twinning on autistic traits. Autism Res. Child Adolesc Psychiatry. 2014;53(2):221-230. microarrays and both community and diagnosed 2011;4(5):377-382. 47. Owens SF, Picchioni MM, Ettinger U, et al. autism replication samples. Behav Genet. 2010;40 (1):31-45. 55. Kendler KS, Eaves LJ. Models for the joint effect Prefrontal deviations in function but not volume are of genotype and environment on liability to putative endophenotypes for schizophrenia. Brain. 51. Steer CD, Golding J, Bolton PF. Traits psychiatric illness. Am J Psychiatry. 1986;143(3): 2012;135(pt 7):2231-2244. contributing to the autistic spectrum. PLoS One. 279-289. 2010;5(9):e12633. doi:10.1371/journal.pone.0012633. 48. Plomin R, DeFries JC, Knopik VS, Neiderhiser 56. Logan JA, Petrill SA, Hart SA, et al. Heritability JM. Behavioural Genetics. 6th ed. New York, NY: 52. St Pourcain B, Skuse DH, Mandy WP, et al. across the distribution: an application of quantile Worth Publishers; 2013. Variability in the common genetic architecture of regression. Behav Genet. 2012;42(2):256-267. social-communication spectrum phenotypes during

jamapsychiatry.com (Reprinted) JAMA Psychiatry May 2015 Volume 72, Number 5 423

Downloaded From: https://jamanetwork.com/ on 10/02/2021